In [1]:
import pandas as pd
In [2]:
import numpy as np
In [3]:
ads = pd.read_csv("train_data_ads.csv")
In [4]:
feeds = pd.read_csv("train_data_feeds.csv")
In [5]:
ads = ads.drop_duplicates(subset='user_id', keep='first', inplace=False)
feeds = feeds.drop_duplicates(subset='u_userId', keep='first', inplace=False)
In [6]:
feeds.shape
Out[6]:
(180123, 28)
In [7]:
ads.shape
Out[7]:
(65297, 35)
In [8]:
feeds['user_id'] = feeds['u_userId']
In [9]:
feeds = feeds.drop('u_userId', axis = 1)
In [10]:
merged = pd.merge(ads, feeds, on = 'user_id', how = 'outer')
In [11]:
merged.head()
Out[11]:
log_id label_x user_id age gender residence city city_rank series_dev series_group ... e_ch e_m e_po e_pl e_rn e_section e_et label_y cillabel pro
0 373250.0 0.0 100005 3.0 2.0 16.0 147.0 2.0 32.0 6.0 ... 19 1123 13 2836 1 0 202206031527 -1 -1 0
1 101100.0 0.0 100006 5.0 2.0 13.0 191.0 4.0 32.0 6.0 ... 19 998 6 835 1 1 202206031125 -1 -1 0
2 742637.0 0.0 100009 5.0 2.0 46.0 354.0 2.0 11.0 8.0 ... 19 508 2 407 3 0 202206080528 -1 -1 0
3 744753.0 0.0 100010 3.0 4.0 33.0 319.0 3.0 31.0 3.0 ... 11 1361 3 2383 3 0 202206032213 -1 -1 0
4 669191.0 0.0 100019 7.0 2.0 16.0 310.0 2.0 16.0 5.0 ... 12 1319 5 2240 2 0 202206032128 -1 -1 0

5 rows × 62 columns

In [12]:
merged.shape
Out[12]:
(180123, 62)
In [13]:
76756 + 32278
Out[13]:
109034
In [14]:
merged.columns
Out[14]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro'],
      dtype='object')
In [15]:
df_cust = merged[merged['label_y'] == 1]
In [16]:
df_cust.shape
Out[16]:
(14450, 62)
In [17]:
df_cust
Out[17]:
log_id label_x user_id age gender residence city city_rank series_dev series_group ... e_ch e_m e_po e_pl e_rn e_section e_et label_y cillabel pro
18 583728.0 0.0 100108 2.0 3.0 21.0 200.0 4.0 30.0 3.0 ... 19 506 20 1100 1 0 202206031522 1 -1 100
21 364370.0 0.0 100127 7.0 2.0 17.0 343.0 5.0 16.0 5.0 ... 19 705 6 2760 1 1 202206080624 1 -1 100
23 588242.0 0.0 100149 8.0 2.0 16.0 425.0 2.0 34.0 7.0 ... 19 928 11 1535 1 0 202206080717 1 -1 100
27 679513.0 0.0 100158 6.0 4.0 33.0 319.0 3.0 27.0 2.0 ... 19 1193 9 332 1 0 202206030115 1 -1 100
29 1084910.0 0.0 100166 5.0 2.0 30.0 113.0 5.0 16.0 5.0 ... 19 1424 6 879 1 1 202206031126 1 -1 100
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
179951 NaN NaN 131907 NaN NaN NaN NaN NaN NaN NaN ... 19 1444 13 1170 1 0 202206051322 1 -1 40
179959 NaN NaN 123724 NaN NaN NaN NaN NaN NaN NaN ... 19 1089 12 2658 1 0 202206051923 1 -1 80
179996 NaN NaN 215157 NaN NaN NaN NaN NaN NaN NaN ... 19 1319 13 2420 1 0 202206050917 1 -1 60
179997 NaN NaN 107610 NaN NaN NaN NaN NaN NaN NaN ... 19 1075 20 505 1 0 202206050816 1 -1 0
180013 NaN NaN 246671 NaN NaN NaN NaN NaN NaN NaN ... 19 620 6 2042 1 1 202206050116 1 -1 0

14450 rows × 62 columns

In [18]:
import matplotlib.pyplot as plt
plt.hist(df_cust['age'])
Out[18]:
(array([ 874., 1744.,  184.,    0., 1673.,  922.,    0., 1589., 1388.,
         204.]),
 array([2. , 2.7, 3.4, 4.1, 4.8, 5.5, 6.2, 6.9, 7.6, 8.3, 9. ]),
 <BarContainer object of 10 artists>)
In [19]:
plt.hist(df_cust['residence'])
Out[19]:
(array([ 651., 1013., 2093.,  353.,  983.,  626., 1544.,   77.,  562.,
         676.]),
 array([11. , 14.5, 18. , 21.5, 25. , 28.5, 32. , 35.5, 39. , 42.5, 46. ]),
 <BarContainer object of 10 artists>)
In [20]:
plt.hist(df_cust['city'])
Out[20]:
(array([ 843.,  888.,  717.,  733.,  621.,  651., 1750., 1079.,  477.,
         819.]),
 array([101., 135., 169., 203., 237., 271., 305., 339., 373., 407., 441.]),
 <BarContainer object of 10 artists>)
In [21]:
plt.hist(df_cust['series_group'])
Out[21]:
(array([1065., 1926.,    0.,  643.,    0., 1705., 1311.,    0., 1153.,
         775.]),
 array([2. , 2.6, 3.2, 3.8, 4.4, 5. , 5.6, 6.2, 6.8, 7.4, 8. ]),
 <BarContainer object of 10 artists>)
In [22]:
plt.hist(df_cust['e_section'])
Out[22]:
(array([8685.,    0.,    0.,    0.,    0.,    0.,    0.,    0.,    0.,
        5765.]),
 array([0. , 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1. ]),
 <BarContainer object of 10 artists>)
In [23]:
import plotly.express as px
fig = px.pie(df_cust, values='age', names='age', title = "Potential Customer Age Distribution")
fig.show()
In [24]:
fig = px.box(df_cust, x="age", title = "Potential Customer Age Distribution (Boxplot)")
fig.show()
In [25]:
import plotly.express as px
fig = px.pie(df_cust, values='residence', names='residence', title = "Potential Customer Residence Distribution")
fig.show()
In [26]:
import plotly.express as px
fig = px.pie(df_cust, values='city', names='city')
fig.show()
In [27]:
merged['e_section'].value_counts()
Out[27]:
1    109898
0     70225
Name: e_section, dtype: int64
In [28]:
df_cust['e_section'].value_counts()
Out[28]:
0    8685
1    5765
Name: e_section, dtype: int64
In [29]:
value_counts = df_cust['e_section'].value_counts()
count_0 = value_counts.get(0, 0)  
count_1 = value_counts.get(1, 0)

labels = ['0', '1']
values = [count_0, count_1]

fig = px.pie(values=values, names=labels, title='Distribution of content preferences among Potential Customers')
fig.show()
In [30]:
df_cust.columns
Out[30]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro'],
      dtype='object')
In [31]:
import plotly.express as px
fig = px.pie(df_cust, values='series_dev', names='series_dev', title = "Potential Customer Device Series Distribution")
fig.show()
In [32]:
import plotly.express as px
fig = px.pie(df_cust, values='series_group', names='series_group', title = "Potential Customer Device Series Group Distribution")
fig.show()
In [33]:
import plotly.express as px
fig = px.pie(df_cust, values='emui_dev', names='emui_dev', title = "Potential Customer Device EMUI Distribution")
fig.show()
In [34]:
import plotly.express as px
fig = px.pie(df_cust, values='device_name', names='device_name', title = "Potential Customer Device Name Distribution")
fig.show()
In [35]:
import plotly.express as px
fig = px.pie(df_cust, values='device_size', names='device_size', title = "Potential Customer Device Size Distribution")
fig.show()
In [36]:
df_cust.columns
Out[36]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro'],
      dtype='object')
In [37]:
df_cust['pt_d'] = pd.to_datetime(df_cust['pt_d'], format='%Y%m%d%H%M')
df_cust['e_et'] = pd.to_datetime(df_cust['e_et'], format='%Y%m%d%H%M')

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\194800331.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\194800331.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [38]:
df_cust['ads_hour'] = df_cust['pt_d'].dt.hour
df_cust['feeds_hour'] = df_cust['e_et'].dt.hour

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\2629954945.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\2629954945.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [39]:
df_cust['ads_day'] = df_cust['pt_d'].dt.dayofweek
df_cust['feeds_day'] = df_cust['e_et'].dt.dayofweek

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1720565557.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1720565557.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [40]:
df_cust['ads_dayname'] = df_cust['pt_d'].dt.day_name()
df_cust['feeds_dayname'] = df_cust['e_et'].dt.day_name()

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1132003763.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1132003763.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [41]:
df_cust.columns
Out[41]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro', 'ads_hour', 'feeds_hour', 'ads_day', 'feeds_day',
       'ads_dayname', 'feeds_dayname'],
      dtype='object')
In [42]:
import plotly.express as px
fig = px.pie(df_cust, values='ads_hour', names='ads_hour', title = "Potential Customer Advertisement Hour Viewed Distribution")
fig.show()
In [92]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['feeds_hour'].value_counts().values, names=df_cust['feeds_hour'].value_counts().index, title = "Potential Customer Feeds Hour Viewed Distribution")
fig.show()
In [44]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['ads_day'].value_counts().values, names=df_cust['ads_dayname'].value_counts().index, title = "Potential Customer Advertisement Day Viewed Distribution")
fig.show()
In [45]:
value_counts = df_cust['e_section'].value_counts()
count_0 = value_counts.get(0, 0)  
count_1 = value_counts.get(1, 0)

labels = ['0', '1']
values = [count_0, count_1]

fig = px.pie(values=values, names=labels, title='Distribution of Content Preferences among Customers')
fig.show()
In [46]:
df_cust['feeds_day']
Out[46]:
18        4
21        2
23        2
27        4
29        4
         ..
179951    6
179959    6
179996    6
179997    6
180013    6
Name: feeds_day, Length: 14450, dtype: int64
In [47]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['feeds_day'].value_counts().values, names=df_cust['feeds_dayname'].value_counts().index, title = "Potential Customer Feeds Day Viewed Distribution")
fig.show()
In [48]:
df_noncust = merged[merged['label_y'] == -1.0]
In [49]:
merged.to_csv("merged_dataframe.csv")
In [50]:
import plotly.express as px
fig = px.pie(df_noncust, values='age', names='age', title = "Non-Potential Customer Age Distribution")
fig.show()
In [51]:
import plotly.express as px
fig = px.pie(df_noncust, values='residence', names='residence', title = "Non-Potential Customer Residence Distribution")
fig.show()
In [52]:
value_counts = df_noncust['e_section'].value_counts()
count_0 = value_counts.get(0, 0)  
count_1 = value_counts.get(1, 0)

labels = ['0', '1']
values = [count_0, count_1]

fig = px.pie(values=values, names=labels, title='Distribution of content preferences among Non-Potential Customers')
fig.show()
In [53]:
import plotly.express as px
fig = px.pie(df_noncust, values='series_group', names='series_group', title = "Non-Potential Customer Device Series Group Distribution")
fig.show()
In [54]:
import plotly.express as px
fig = px.pie(df_noncust, values='series_dev', names='series_dev', title = "Non-Potential Customer Device Series Distribution")
fig.show()
In [55]:
import plotly.express as px
fig = px.pie(df_noncust, values='emui_dev', names='emui_dev', title = "Non-Potential Customer Device EMUI Distribution")
fig.show()
In [56]:
df_noncust['pt_d'] = pd.to_datetime(df_noncust['pt_d'], format='%Y%m%d%H%M')
df_noncust['e_et'] = pd.to_datetime(df_noncust['e_et'], format='%Y%m%d%H%M')

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\372196138.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\372196138.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [57]:
df_noncust['ads_hour'] = df_noncust['pt_d'].dt.hour
df_noncust['feeds_hour'] = df_noncust['e_et'].dt.hour

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\137544654.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\137544654.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [58]:
df_noncust['ads_day'] = df_noncust['pt_d'].dt.dayofweek
df_noncust['feeds_day'] = df_noncust['e_et'].dt.dayofweek

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1650693043.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\1650693043.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [59]:
df_noncust['ads_dayname'] = df_noncust['pt_d'].dt.day_name()
df_noncust['feeds_dayname'] = df_noncust['e_et'].dt.day_name()

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\4293718302.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\anime\AppData\Local\Temp\ipykernel_10964\4293718302.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
In [60]:
import plotly.express as px
fig = px.pie(df_noncust, values='ads_hour', names='ads_hour', title = "Non-Potential Customer Advertisement Hour Viewed Distribution")
fig.show()
In [93]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['feeds_hour'].value_counts().values, names=df_noncust['feeds_hour'].value_counts().index, title = "Non-Potential Customer Feeds Hour Viewed Distribution")
fig.show()
In [62]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['ads_day'].value_counts().values, names=df_noncust['ads_dayname'].value_counts().index, title = "Non-Potential Customer Advertisement Day Viewed Distribution")
fig.show()
In [63]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['feeds_day'].value_counts().values, names=df_noncust['feeds_dayname'].value_counts().index, title = "Non-Potential Customer Feeds Day Viewed Distribution")
fig.show()
In [64]:
df_cust.to_csv("customer_df.csv")
In [65]:
df_cust.columns
Out[65]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro', 'ads_hour', 'feeds_hour', 'ads_day', 'feeds_day',
       'ads_dayname', 'feeds_dayname'],
      dtype='object')
In [66]:
df_noncust.to_csv("noncustomer_df.csv")
In [ ]:
In [67]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['age'].value_counts().index, y=df_cust['age'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['age'].value_counts().index, y=df_noncust['age'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Age Distribution Differences")
fig.show()
In [68]:
fig = px.pie(merged, values=merged['label_y'].value_counts().values, names=merged['label_y'].value_counts().index, title = "Label Distribution")
fig.show()
In [69]:
merged.shape
Out[69]:
(180123, 62)
In [70]:
merged['label_y'].value_counts()
Out[70]:
-1    165673
 1     14450
Name: label_y, dtype: int64
In [71]:
merged.columns
Out[71]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro'],
      dtype='object')
In [72]:
ads.columns
Out[72]:
Index(['log_id', 'label', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST', 'u_refreshTimes', 'u_feedLifeCycle'],
      dtype='object')
In [73]:
feeds.columns
Out[73]:
Index(['u_phonePrice', 'u_browserLifeCycle', 'u_browserMode',
       'u_feedLifeCycle', 'u_refreshTimes', 'u_newsCatInterests',
       'u_newsCatDislike', 'u_newsCatInterestsST', 'u_click_ca2_news',
       'i_docId', 'i_s_sourceId', 'i_regionEntity', 'i_cat', 'i_entities',
       'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch', 'e_m', 'e_po', 'e_pl',
       'e_rn', 'e_section', 'e_et', 'label', 'cillabel', 'pro', 'user_id'],
      dtype='object')
In [74]:
feeds.shape
Out[74]:
(180123, 28)
In [75]:
feeds['cillabel'].value_counts()
Out[75]:
-1    180046
 1        77
Name: cillabel, dtype: int64
In [76]:
import plotly.express as px
fig = px.pie(df_cust, values='gender', names='gender', title = "Potential Customer Gender Distribution")
fig.show()
In [77]:
import plotly.express as px
fig = px.pie(df_noncust, values='gender', names='gender', title = "Non Customer Gender Distribution")
fig.show()
In [78]:
df_cust['gender'].unique()
Out[78]:
array([ 3.,  2.,  4., nan])
In [79]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['gender'].value_counts().index, y=df_cust['gender'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['gender'].value_counts().index, y=df_noncust['gender'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Age Distribution Differences")
fig.show()
In [80]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['residence'].value_counts().index, y=df_cust['residence'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['residence'].value_counts().index, y=df_noncust['residence'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Residence Distribution Differences")
fig.show()
In [81]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['e_section'].value_counts().index, y=df_cust['e_section'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['e_section'].value_counts().index, y=df_noncust['e_section'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Content Preference Distribution Differences")
fig.show()
In [82]:
df_cust['e_section'].value_counts()
Out[82]:
0    8685
1    5765
Name: e_section, dtype: int64
In [83]:
df_noncust['e_section'].value_counts()
Out[83]:
1    104133
0     61540
Name: e_section, dtype: int64
In [84]:
df_cust.columns
Out[84]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro', 'ads_hour', 'feeds_hour', 'ads_day', 'feeds_day',
       'ads_dayname', 'feeds_dayname'],
      dtype='object')
In [85]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['series_dev'].value_counts().index, y=df_cust['series_dev'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['series_dev'].value_counts().index, y=df_noncust['series_dev'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Device Series Distribution Differences")
fig.show()
In [86]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['series_group'].value_counts().index, y=df_cust['series_group'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['series_group'].value_counts().index, y=df_noncust['series_group'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Device Series Group Distribution Differences")
fig.show()
In [87]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['emui_dev'].value_counts().index, y=df_cust['emui_dev'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['emui_dev'].value_counts().index, y=df_noncust['emui_dev'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Device EMUI Distribution Differences")
fig.show()
In [88]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['ads_hour'].value_counts().index, y=df_cust['ads_hour'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['ads_hour'].value_counts().index, y=df_noncust['ads_hour'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Ad Hour Viewed Distribution Differences")
fig.show()
In [89]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['feeds_hour'].value_counts().index, y=df_cust['feeds_hour'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['feeds_hour'].value_counts().index, y=df_noncust['feeds_hour'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Feed Hour Viewed Distribution Differences")
fig.show()
In [90]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['feeds_day'].value_counts().index, y=df_cust['feeds_day'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['feeds_day'].value_counts().index, y=df_noncust['feeds_day'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Feed Day Viewed Distribution Differences")
fig.show()
In [91]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['ads_day'].value_counts().index, y=df_cust['ads_day'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['ads_day'].value_counts().index, y=df_noncust['ads_day'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Ad Day Viewed Distribution Differences")
fig.show()
In [94]:
df_cust.columns
Out[94]:
Index(['log_id', 'label_x', 'user_id', 'age', 'gender', 'residence', 'city',
       'city_rank', 'series_dev', 'series_group', 'emui_dev', 'device_name',
       'device_size', 'net_type', 'task_id', 'adv_id', 'creat_type_cd',
       'adv_prim_id', 'inter_type_cd', 'slot_id', 'site_id', 'spread_app_id',
       'hispace_app_tags', 'app_second_class', 'app_score',
       'ad_click_list_v001', 'ad_click_list_v002', 'ad_click_list_v003',
       'ad_close_list_v001', 'ad_close_list_v002', 'ad_close_list_v003',
       'pt_d', 'u_newsCatInterestsST_x', 'u_refreshTimes_x',
       'u_feedLifeCycle_x', 'u_phonePrice', 'u_browserLifeCycle',
       'u_browserMode', 'u_feedLifeCycle_y', 'u_refreshTimes_y',
       'u_newsCatInterests', 'u_newsCatDislike', 'u_newsCatInterestsST_y',
       'u_click_ca2_news', 'i_docId', 'i_s_sourceId', 'i_regionEntity',
       'i_cat', 'i_entities', 'i_dislikeTimes', 'i_upTimes', 'i_dtype', 'e_ch',
       'e_m', 'e_po', 'e_pl', 'e_rn', 'e_section', 'e_et', 'label_y',
       'cillabel', 'pro', 'ads_hour', 'feeds_hour', 'ads_day', 'feeds_day',
       'ads_dayname', 'feeds_dayname'],
      dtype='object')
In [95]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['i_upTimes'].value_counts().values, names=df_cust['i_upTimes'].value_counts().index, title = "Potential Customer Article Like Distribution")
fig.show()
In [96]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['i_upTimes'].value_counts().values, names=df_noncust['i_upTimes'].value_counts().index, title = "Non-Potential Customer Article Like Distribution")
fig.show()
In [97]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['i_dislikeTimes'].value_counts().values, names=df_cust['i_dislikeTimes'].value_counts().index, title = "Potential Customer Article Dislike Distribution")
fig.show()
In [98]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['i_dislikeTimes'].value_counts().values, names=df_noncust['i_dislikeTimes'].value_counts().index, title = "Non-Potential Customer Article Dislike Distribution")
fig.show()
In [99]:
import plotly.express as px
fig = px.pie(df_cust, values= df_cust['pro'].value_counts().values, names=df_cust['pro'].value_counts().index, title = "Potential Customer Article Progress Distribution")
fig.show()
In [101]:
df_noncust['pro'].value_counts()
Out[101]:
0    165673
Name: pro, dtype: int64
In [100]:
import plotly.express as px
fig = px.pie(df_noncust, values= df_noncust['pro'].value_counts().values, names=df_noncust['pro'].value_counts().index, title = "Non-Potential Customer Article Progress Distribution")
fig.show()
In [102]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['i_upTimes'].value_counts().index, y=df_cust['i_upTimes'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['i_upTimes'].value_counts().index, y=df_noncust['i_upTimes'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Article Liked Distribution Differences")
fig.show()
In [103]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['i_dislikeTimes'].value_counts().index, y=df_cust['i_dislikeTimes'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['i_dislikeTimes'].value_counts().index, y=df_noncust['i_dislikeTimes'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Article Disliked Distribution Differences")
fig.show()
In [104]:
import plotly.graph_objects as go


fig = go.Figure(data=[
    go.Bar(name='Potential Customers', x=df_cust['pro'].value_counts().index, y=df_cust['pro'].value_counts().values),
    go.Bar(name='Non Customers', x=df_noncust['pro'].value_counts().index, y=df_noncust['pro'].value_counts().values)
])
# Change the bar mode
fig.update_layout(barmode='group', title = "Grouped Barchart to Visualize Article Progress Distribution Differences")
fig.show()
In [ ]: